Disclosure of Invention
The invention aims to overcome the defects of the prior art and provide a method for hierarchical planning of an offshore wind farm annular current collection network.
The purpose of the invention can be realized by the following technical scheme:
a hierarchical planning method for an offshore wind farm annular collector network comprises the following steps:
s1: reading in offshore wind farm data, wherein the wind farm data comprises a fan coordinate, an offshore substation coordinate, equipment cost and parameters;
s2: presetting an optimal partition number, and obtaining an initial clustering center and a partition division matrix of each partition;
s3: performing preliminary string optimization on each partition to obtain the total cost of submarine cables on two sides of each node of all initial strings, the length of each high-voltage cable of each string and the Euclidean distance from each node to a clustering center, wherein the node is a fan;
s4: updating a clustering center and a partition matrix according to the total cost of submarine cables on two sides of each node of all sub-strings, the length of each high-voltage cable of each sub-string and the Euclidean distance from each node to the clustering center;
s5: performing annular structure string optimization on each partition by adopting a single-parent genetic algorithm, and updating the total cost of submarine cables on two sides of each node of all strings, the length of each high-voltage cable of each string and the Euclidean distance from each node to a clustering center;
s6: and calculating an optimization value according to the total cost of the submarine cables on two sides of each node of all the sub-strings, the length of each high-voltage cable of each sub-string and the Euclidean distance from each node to the clustering center in combination with the current clustering center and the partition division matrix, judging whether the difference value between the current optimization value and the previous optimization value is smaller than a threshold value, if not, returning to the step S4, and if so, finishing the optimization to obtain the planning result of the annular collector network.
The element μ in the partition dividing matrix updated in the step S4ijThe method specifically comprises the following steps:
CHij=CHi/n
CHlj=CHl/n
wherein: mu.sijDegree of membership, d, of node j to clustering center iijIs the Euclidean distance, λ, from node j to cluster center im、λhTo adjust the coefficients, CMijThe total cost CH of submarine cables on two sides of the node j in the partition corresponding to the clustering center iij、CHljLength equivalent to the length of the high-voltage cable, CHiIs the length of the ith section of high-voltage cable, n is the number of nodes, dljIs the Euclidean distance, CM, from node j to cluster center lljFor the total cost, CH, of the submarine cables on both sides of node j in partition llThe length of the first section of high-voltage cable, c is the number of partitions, and m is a weighting index;
the clustering center is specifically as follows:
wherein: p is a radical ofiAs the clustering center i, xjFor node j, pc is the coordinate vector of the point of common connection.
The optimization degree value Obj is:
CHij=CHi/n
wherein: mu.sijIs the degree of membership of the node j to the clustering center i, c is the number of partitions, n is the number of nodes, dijIs the Euclidean distance, λ, from node j to cluster center im、λhTo adjust the coefficients, CMijThe total cost CH of submarine cables on two sides of the node j in the partition corresponding to the clustering center iijLength equivalent to the length of the high-voltage cable, CHiIs the length of the ith high-voltage cable.
Length CH of the ith segment of high-voltage cableiAnd taking the distance from the clustering center i to the onshore common connection point.
The step S5 specifically includes the steps of:
s51: performing annular structure string optimization on each partition by adopting a single-parent genetic algorithm;
s52: and performing cable type selection, and updating the total cost of submarine cables on two sides of each node of all sub-strings, the length of each high-voltage cable of each sub-string and the Euclidean distance from each node to a clustering center.
Compared with the prior art, the invention has the following advantages:
1) the method adopts a layered planning optimization mode of a transformer substation layer, a fan layer and a cable layer, simplifies the complexity of optimization, and can reduce the cost and improve the reliability.
2) The optimization of the clustering center and the partition matrix is obtained based on the total cost of submarine cables on two sides of each node of all the sub-strings, the length of each high-voltage cable of each sub-string and the Euclidean distance from each node to the clustering center, and the cost-oriented optimization can be realized.
3) The optimization value takes the minimum sum of the distortion of each sample point and the clustering center, namely the Euclidean distance between two vectors and the submarine cable length after the partition topology optimization as an objective function, so that the optimization effect on the submarine cable can be improved.
4) And the distance from the clustering center i to the onshore common connection point is taken, so that the calculation can be simplified while the accuracy is ensured.
Detailed Description
The invention is described in detail below with reference to the figures and specific embodiments. The present embodiment is implemented on the premise of the technical solution of the present invention, and a detailed implementation manner and a specific operation process are given, but the scope of the present invention is not limited to the following embodiments.
Firstly, a mathematical model of the wind farm current collection network needs to be established
Factors that affect the cost of the current collection network of an offshore wind farm include: the number, the position and the voltage grade of the wind turbines, the load of electrical equipment, the topological wiring form, the cable section, the length, the arrangement of a cable trench and the like. The variables are correlated and constrained with each other, so that the current collection network optimization model has the characteristics of multiple discrete variables and strong nonlinearity. The final aim of the topological structure optimization of the AC current collection network of the large offshore wind farm is to strive for the lowest cost of the current collection network on the basis of meeting the reliability requirement. The large-scale offshore wind farm current collection network mainly comprises a submarine cable current collection network connected with each fan, an offshore substation and a transmission cable connected with a continental power grid, and has the characteristics of multiple voltage levels and multiple current collection subregions on the whole.
Fig. 1 shows a schematic diagram of a current collection system topology.
Cost of current collection network Cwhole_csThe method mainly comprises the cost of a tower bottom box transformer of the wind generating set, the cost of a medium-voltage submarine cable of a current collection network, the cost of electrical equipment (a transformer and other auxiliary electrical equipment) of an offshore substation and the cost of a high-voltage transmission cable, and is shown as the following formula:
Cwhole_cs=CWTtran+CSUB+CCable (1)
wherein, CSUBFor offshore transformer substationTotal cost C of wind generating set and tower bottom box transformer substationWTtranThe quantization can be:
CWTtran=Nwt·PWTtran
wherein N iswtThe number of the fans in the large-scale offshore wind farm, PWTtranIs the unit price of the box transformer substation.
Total cost of cable CCableIncluding the total cost C of medium voltage sea cableMV_cableAnd total cost of high voltage cable CHV_cableTwo parts. Since the current collection network structure is coupled with the power flow and short circuit current distribution, the nonlinear influence of the discrete variables on each other must be effectively described in the modeling process.
CCable=CHV_cable+CMV_cable (2)
Wherein N iss、NsfAnd NsfcThe number of the offshore substations in the large offshore wind farm, the number of the feeders (namely, the number of the fan strings) connected with the s-th offshore substation, and the number of the sections of the submarine cables in the f-th feeder of the s-th substation are respectively. Ccable,sfcFor corresponding sea cable costs (including installation and transportation costs C of sea cable)install、Ctrans)。
Ccable,sfc=Lsfc·Cunit(typei) (5)
Wherein L issfcIs the length of the sea cable at the C-th section in the f-th feeder line on the s-th substation, Cunit(typei) For the unit price (type) of the sea cablei)。Cunit(typei) The maximum allowable voltage drop Δ V is passed through the load currentmax(related to the length of each section of sea cable), maximum short circuit current, etc. L ishvcable,sIs the s th seaA length of high voltage cable between the substation and the point of common connection.
The cost of electrical equipment in offshore substations mainly includes the main transformation cost of each offshore substation and the cost of main transformation auxiliary equipment such as protection equipment and control equipment.
Wherein P isOStran,sThe unit price of the main transformer of the s-th offshore substation; n is a radical ofOStran,sThe number of main transformer stations of the s offshore substation is shown; celec,sFor the cost of other electrical equipment of the s-th offshore substation, high and medium voltage GIS switches are mainly considered herein.
Furthermore, offshore wind farms also need to take into account specific sea conditions. On one hand, the laying of the submarine cables cannot be crossed and cannot be in the same ditch, and the crossing with other pipelines in the sea area, the channel constraint and the like are required to be reduced as much as possible; on the other hand, the scheme design needs to meet the optimality and feasibility of the calculation result.
In summary, the mathematical model of the topological structure optimization design of the ring-shaped current collection network of the large offshore wind farm is described as follows:
ST:Isfc<Irated(typei)
|ΔVsfc|<ΔVmax
wherein, Isfc、ΔVsfc is the current flowing on the c-th section of submarine cable in the f-th feeder line of the s-th transformer substation and the voltage drop of the section of submarine cable。FiThe wind turbine nodes are a fan node set, and since submarine cables cannot be laid in a crossed mode, constraint conditions require that no intersection exists among fan clusters, and all fans are contained in the fan clusters.
The rated capacity of a large-scale offshore wind farm is generally hundreds of gigawatts or even gigawatts, under the condition, one offshore substation is difficult to meet the power transformation requirement, in addition, the number of fans of the wind farm on the scale is large, the distribution range is wide, the cost and the difficulty of laying medium-voltage submarine cables are undoubtedly increased by only arranging one offshore substation, and meanwhile, the reliability of the whole wind farm is reduced. Therefore, the division of the power transformation area of the large offshore wind farm is very necessary.
The fuzzy clustering algorithm based on the objective function is adopted for partitioning (namely partitioning) the current collecting sub-region of the wind power plant. A Fuzzy Clustering Method (FCM for short) based on an objective function is to classify clusters into a nonlinear programming problem with constraints, and obtain Fuzzy partition and cluster partition of a data set through optimization solution. The method comprises the following steps of taking coordinates of all wind generating sets in a large offshore wind farm as a data set, taking the number of offshore substations to be planned as a clustering number, and taking the distortion degree of each sample point and a clustering center, namely the minimum sum of the Euclidean distance between two vectors and the sea cable length after partition topology optimization, as an objective function to perform clustering calculation, wherein the method specifically comprises the following steps:
solving method of current collection network optimization model
Fuzzy clustering algorithm partition based on target function
St:U∈Mfc
Wherein: n is the number of nodes, namely the number of fans; c is the number of clusters (i.e., the number of Shanghai substations); mu.sijFor dividing the elements of the matrix U, representing the element j to the cluster centre iDegree of membership; dijThe Euclidean distance from node j to cluster center i is expressed as follows:
(dij)2=|xj-pi|A=(xj-pi)TA(xj-pi) (9)
wherein A is an identity matrix; CM (compact message processor)
ijThe total cost of submarine cables on two sides of the node j in the annular group i; CH (CH)
ijLength of equivalent high voltage cable: CH (CH)
ij=CH
i/n,
Wherein CH
iThe length of the ith section of high-voltage cable is approximately equal to the distance from the offshore substation to the onshore common connection point in the offshore subarea:
CHi=||pi-pcc||·Cunit(typei) (10)
where pc is the coordinate vector of the point of common connection. Lambda [ alpha ]m、λhTo adjust the coefficient, a constant is used to adjust the cost data and the distance data to the same order of magnitude.
The constraint of the above objective function is the equation:
the lagrange function obtained by the lagrange condition extremum method is as follows:
l to λ and μijRespectively obtaining derivatives, and finally obtaining:
the same principle is that:
order:
then there are:
solving a partition matrix U and a clustering center piAnd finally obtaining the optimal number of the subareas, namely the planning number of the offshore transformer substation, and the clustering center in the subareas, namely the position of the offshore transformer substation in the subareas through iterative calculation, wherein the subarea to which each wind generating set belongs can be known through a partition matrix.
A method for hierarchical planning of an offshore wind farm annular collector network is disclosed, as shown in FIG. 2, and comprises the following steps:
s1: reading in offshore wind farm data, wherein the wind farm data comprises a fan coordinate, an offshore substation coordinate (if known), equipment cost and parameters;
s2: presetting an optimal partition number (which needs to be determined when the partition number is unknown), and obtaining an initial clustering center and a partition division matrix of each partition;
s3: performing preliminary string optimization on each partition to obtain the total cost of submarine cables on two sides of each node of all initial strings, the length of each high-voltage cable of each string and the Euclidean distance from each node to a clustering center, wherein the node is a fan;
s4: updating a clustering center and a partition matrix according to the total cost of submarine cables on two sides of each node of all the current sub-strings, the length of each high-voltage cable of each sub-string and the Euclidean distance from each node to the clustering center, wherein an element mu in the updated partition matrixijThe method specifically comprises the following steps:
CHij=CHi/n
CHlj=CHl/n
wherein: mu.sijDegree of membership, d, of node j to clustering center iijIs the Euclidean distance, λ, from node j to cluster center im、λhTo adjust the coefficients, CMijThe total cost CH of submarine cables on two sides of the node j in the partition corresponding to the clustering center iij、CHljLength equivalent to the length of the high-voltage cable, CHiIs the length of the ith high-voltage cable (i.e. the ith zone to the shore entry point), n is the number of nodes, dljIs the Euclidean distance, CM, from node j to cluster center lljFor the total cost, CH, of the submarine cables on both sides of node j in partition llThe length of the first section of high-voltage cable, c is the number of partitions, and m is a weighting index;
the clustering center is specifically as follows:
wherein: p is a radical ofiAs the clustering center i, xjFor node j, pc is the coordinate vector of the point of common connection.
Length CH of ith section of high voltage cableiAnd taking the distance from the clustering center i to the onshore common connection point.
S5: adopting a single-parent genetic algorithm to carry out annular structure string optimization on each partition, updating the total cost of submarine cables on two sides of each node of all strings, the length of each high-voltage cable of each string and the Euclidean distance from each node to a clustering center, and specifically comprising the following steps:
s51: performing annular structure string optimization on each partition by adopting a single-parent genetic algorithm; the single parent genetic algorithm is divided into strings:
after the topological structure of the large offshore wind power plant is planned and partitioned, a Single Parent Genetic Algorithm (SPGA for short) is adopted to optimize the topological structure of a certain partition. Compared with the traditional genetic algorithm, the SPGA has the outstanding characteristics that each daughter only has one parent in the generation process of the daughter population, and new individuals with different characters are generated by randomly executing transposition operators or reversal operators on the parents. The single-parent transposition operator can ensure that a new generation of individuals has the basic characteristic of becoming a feasible solution and can also improve the searching capacity of a solution space. The method has the advantages that the execution speed of the single-parent reverse operator is high, the effective gene segments in the parent can be directly inherited into the daughter, one chromosome carries path information (namely, a path chromosome) and the other chromosome carries grouping information (namely, a breakpoint chromosome) by adopting a parallel inheritance mode of a chromosome pair, and the two chromosomes adopt simple and intuitive integer codes, as shown in figure 3.
One chromosome pair represents a topology of each subset of the electrical networks in the partition, i.e., the multi-traveler path that characterizes the submarine cable path, so that the number of offshore wind turbines connected to each section of submarine cable and the maximum short-circuit current flowing through can be obtained. And (4) carrying out submarine cable type selection and verification according to the current-carrying capacity and the maximum short-circuit current of each section of submarine cable. And obtaining the manufacturing cost of the medium-voltage submarine cable according to the selected type of the submarine cable, the corresponding unit price of the submarine cable and the length of the submarine cable. The cost of the medium-voltage submarine cable is added with the cost of other equipment in the subset electric network, such as a box transformer substation, an offshore substation and the like, so that the cost of the subset electric network topology mode corresponding to the chromosome is obtained and is used as the fitness value of the chromosome, and the specific calculation method is analyzed in the optimization model.
Considering the special constraint that submarine cable connection can not be crossed, a penalty function positively correlated with the crossing times is added into the fitness function. By setting corresponding parameter values, the situation of submarine cable crossing does not occur in the final optimization scheme.
S52: and performing cable type selection, and updating the total cost of submarine cables on two sides of each node of all sub-strings, the length of each high-voltage cable of each sub-string and the Euclidean distance from each node to a clustering center. And simultaneously, selecting the type of the cable, calculating the total cost of the medium-voltage cable, and adding the total cost of the medium-voltage cable, the box transformer substation cost and the offshore substation cost to obtain the total cost of the subarea current collection system, namely the individual adaptability value.
Routing connection of multi-traveler problem: the multi-traveler Problem (MTSP) is an extension of the traveler Problem, that is, there are N cities, and M travelers respectively start from the same city (or different cities), and walk a travel route to visit all cities, so that each city has only one traveler passing through (except for the starting city), and finally each traveler returns to the original starting city (forming a circular path), and the total travel distance is shortest. By taking the solution of the problems of multiple traveling salesmen as a reference, an offshore wind farm partition containing N wind generating sets is regarded as a set of N cities, and the wind generating sets in the partition are divided into M annular wind generating set strings. Therefore, the design problem of the ring current collection network topology of the offshore wind farm is converted into the MTSP problem that M travelers visit N cities. In addition, according to the number of cities walked by each traveling salesman, whether the tasks of the MTSP are equally divided is judged, namely whether the number of the connected fans in each string is necessarily equal or close. The route description for the multi-traveler problem is shown in fig. 4.
S6: and calculating an optimization value according to the total cost of the submarine cables on two sides of each node of all the sub-strings, the length of each high-voltage cable of each sub-string and the Euclidean distance from each node to the clustering center in combination with the current clustering center and the partition division matrix, judging whether the difference value between the current optimization value and the previous optimization value is smaller than a threshold value, if not, returning to the step S4, and if so, finishing the optimization to obtain the planning result of the annular collector network.
The optimization value Obj is:
CHij=CHi/n
wherein: mu.sijIs the degree of membership of the node j to the clustering center i, c is the number of partitions, n is the number of nodes, dijIs the Euclidean distance, λ, from node j to cluster center im、λhTo adjust the coefficients, CMijThe total cost CH of submarine cables on two sides of the node j in the partition corresponding to the clustering center iijLength equivalent to the length of the high-voltage cable, CHiIs the length of the ith high-voltage cable.
Examples of specific applications
The case wind power plant is a large offshore wind power plant which is 50km offshore and comprises 259 wind power generating sets with single machine capacity of 3.5MW, and specific parameters of the offshore wind power generating sets are shown in a table 1. Coordinates of the offshore wind generating set are determined, and the wind power station is arranged as shown in the figure 5. By integrating the factors of equipment cost and transmission loss, it is generally considered that 30-36 kV is the optimal voltage level for connection between fans in an alternating current electrical system. The power collection network is connected by submarine medium-voltage cables, the voltage level is 35kV, the power transmission cables are high-voltage submarine cables, and the voltage level is 110kV or 220 kV.
TABLE 1 Individual offshore wind turbine parameters
On the basis of considering factors such as the number of feasible subareas of a large offshore wind power plant, the number of clusters of offshore wind generating sets which are in feasible annular connection in the subareas, the type selection of collecting network equipment and the like, the optimal design method for the collecting network with the annular structure provided by the invention is adopted to optimally design the collecting network of the actual offshore wind power plant in a plan. The main parameters of the 35kV submarine cable are shown in table 2, according to the current-carrying capacity and short-circuit characteristics of the submarine cable, the submarine cable with a 400 section can allow 9 fans of 3.5MW to flow simultaneously, if the number exceeds this, no thicker submarine cable is matched with the submarine cable, and the submarine cable cannot be selected. Therefore, the number of fans per cluster is optimized between 1 and 9. The fewer the number of the fans in each cluster, the higher the reliability, but the huge economic cost; the more the number of the fans in each series is, the higher the economical efficiency is, and certain reliability requirements need to be met.
TABLE 2 Main parameters of Medium Voltage (35kV) submarine Cable
As can be seen from table 2, the medium voltage submarine cable is expensive, the laying length of the medium voltage submarine cable inside the 100MW offshore wind farm of the east sea bridge in the upper sea is about 70km, and if the capacity of the offshore wind farm is larger, the medium voltage submarine cable has a larger optimized space. The method takes the economic optimization as a target, takes the reliability as a constraint, divides a large-scale offshore current collection system into a transformer substation layer, a fan layer and a submarine cable layer according to the characteristics of the large-scale offshore current collection system under the condition of meeting the requirements of special offshore conditions, and adopts the algorithm to respectively optimize and solve each layer. The number of the subareas is optimized at a substation level (the number of the fans in each subarea is limited by the current-carrying capacity and the short-circuit characteristic of a high-voltage submarine cable), the number of the subareas is optimized at a fan level (the number of the fans in each subarea is limited by the current-carrying capacity and the short-circuit characteristic of a medium-voltage submarine cable), and the cross section and the length of the submarine cable are optimized at a submarine cable layer. Two optimization design schemes with the partition numbers of 3 and 4 are finally obtained, specifically shown in fig. 6, and the summary results of the costs of the electrical equipment in the optimization schemes are given, and are shown in tables 2 and 3.
TABLE 2 summary of substation equipment usage and cost
TABLE 3 summary of medium voltage submarine cables of different cross-sections
For comparison, the case wind farm is also designed with a radial topology. The radial structure topology adopts the minimum spanning tree algorithm based on graph theory which is widely used at present, and the design result under the same partition number is shown in FIG. 7. The two types of typical connection structures are used for quantitatively evaluating the optimization results in both economic and reliability aspects. The total cost of the current collection network for both configurations is shown in table 4.
TABLE 4 Total cost of the current-collecting network under different optimization algorithms
According to the existing power generation reliability evaluation experience, under the conditions of lacking power generation system reliability data, load prediction data and insufficient experience, a probability evaluation method is usually selected to be ideal. The offshore wind farm power collection system has numerous and various types of equipment, and complex state combination and operation conditions, and the reliability of the power collection system is estimated by adopting a Monte Carlo simulation method in consideration of the availability [19] of the offshore wind power generation set. The selection of the wind generating sets, the cable failure rate and the repair time are shown in table 5, and the time chart of the number of the available wind generating sets in the collection network with the radial and annular structure with the optimal economy in one year is shown in fig. 8.
TABLE 5 Fault parameters of offshore wind turbines and cables
The available fan capacity for a power collection network of both configurations over the year is shown in fig. 8. From an economic point of view, the cost of the annular collector network is increased by 24 percent compared with that of the radial structure collector network, but the average available capacity of the annular collector network is also increased by 23.7 percent. The cost difference between the two is mainly in the cost of the cable. The reliability of the annular structure wind power plant is obviously superior to that of the radial structure wind power plant, when the offshore wind power plant adopts the annular structure collecting network, the available capacity of the offshore wind power plant is basically equal to the rated total capacity, the stable and efficient electric energy output of the wind power plant is ensured, and when the offshore wind power plant adopts the radial structure collecting network, the available capacity of the offshore wind power plant has large variation amplitude and lower average capacity. Although the economic cost of the ring structure is larger than that of the radial structure, the improvement of the reliability is obvious, so that the ring structure is not lost as an advantageous choice for the wind power plant with strict reliability requirements.
The emphasis here is on comparing the merits of different topologies, the average available capacity in fig. 8 does not account for the wind energy effect, and the generated energy yield value accounting for the wind energy effect is shown in table 6. Annual hours of use are affected by offshore wind energy, but wind energy affects the same for the same offshore wind farm, so the reference 2600 hours of the offshore wind farm design of the above-sea east-sea bridge is used herein as a reference standard. The power generation output value is calculated according to the condition that the power price of the offshore wind power project which is put into operation before 2017 and is determined by the Chinese national development and reform Commission is 0.85 yuan/kWh. A comprehensive comparison of the different topologies is shown in table 6.
TABLE 6 comprehensive comparison of different topologies
From table 6, it can be known that although the investment cost of the annular structure is higher in the initial construction period, the power generation output value of the wind power plant in the 20-year operation period is far higher than that of the radial structure due to the higher reliability of the annular structure. The data in table 6 only consider the reliability factor of the topology structure, but do not consider the problem of difficult maintenance of the offshore wind farm due to poor accessibility, and if the accessibility influence of the offshore wind farm is considered, the benefits brought by the ring structure are more huge.